A two transistor cmos inverter has the two transistor gates coupled together by a coupling capacitor. D-C gate bias is supplied to each transistor through high value resistors. The P-channel transistor is biased one threshold below VDD and the N-channel transistor is biased one threshold above ground. The biasing voltages are developed through the use of a current mirror so that the biasing is independent of processing variables and temperature. This form of biasing renders the circuit class B regardless of the source to drain voltage and ensures low current operation. A crystal oscillator created using such an inverter and biasing will operate at voltages substantially below sum of P and N thresholds and at a current level about one-fifth of that of a conventional cmos oscillator.
|
1. A cmos oscillator circuit comprising:
first and second terminals connectable to a source of operating power; first and second complementary transistors having their source drain circuits directly connected in series across said first and second terminals; capacitive means coupling the gates of said first and second transistors together; resonant circuit means coupled between the drain and gate of one of said pair of transistors to create an oscillatory feedback; a current mirror configured to develop a threshold biasing potential; high value resistor means for coupling said threshold biasing potential to the gate of at least one of said first and second transistors; and a buffer coupled between said current mirror and said one of said first and second transistors.
|
The invention relates to complementary metal oxide semiconductor (CMOS) devices, and in particular, relates to a low power type of oscillator. Normally, an oscillator is created using a CMOS inverter by coupling a resonant circuit between the output and input terminals. Such oscillators draw very little current and operate best when the applied voltage is equal to the sum of the P-channel and N-channel device threshold voltages. Once the voltage is reduced below this value the circuit performance deteriorates rapidly. Oscillator startup will not occur for supply-voltage values below the sum of device thresholds. While the circuit operates in class B at sum of thresholds, when the supply voltage exceeds the value, switching transients become a problem and the circuit will draw appreciable current in oscillator operation. While this current is low and while CMOS circuitry has gained widespread use in battery-operated devices, it would be desirable to reduce the current drain.
U.S. Pat. No. 4,211,985 shows one approach to current drain reduction. High value resistance elements are coupled in series with the drain electrodes in a CMOS inverter. Each transistor is biased by returning its gate through a high value resistance to its drain. This forces each transistor to operate at its threshold. To operate effectively, this oscillator is coupled to a second CMOS inverter pair which is biased from the same resistors, but has its collectors commonly connected.
It is an object of the invention to create a CMOS oscillator in which the two complementary transistors are optimally biased class B from an external bias source.
It is a further object of the invention to employ a CMOS current mirror to develop a first voltage one threshold above ground and a second voltage that is one threshold below the supply voltage and to use these two voltages to bias the two elements in a CMOS inverter to produce class B operation.
It is a still further object of the invention to bias a CMOS inverter in an oscillator from an external source which provides optimum starting of oscillation and operates at reduced current drain.
These and other objects are achieved as follows. A CMOS inverter is created with a capacitor coupling the two gates together for a-c. A CMOS current mirror is used to develop a first voltage that is one N-channel threshold above ground and a second voltage that is one P-channel threshold below VDD. A pair of large value resistors is employed to couple these bias voltages to the inverter N-channel and P-channel device gates respectively. When a tuned circuit is coupled between the inverter output and one of the gates, an oscillator is created which is automatically class B biased regardless of the VDD value. Such an oscillator will start up and run at voltage levels substantially lower than those required by the prior art circuits and less current is required in operation.
In one alternative embodiment only the P-channel device is biased from a current mirror while the N-channel is biased from the normally used feedback resistor. In another alternative embodiment, a pair of transient suppressing resistors is included in the inverter and an isolation buffer is coupled between the current mirror and the P-channel device biasing resistor.
FIG. 1 is a schematic diagram of a prior art oscillator.
FIG. 2 is a schematic diagram of the circuit of the invention.
FIG. 3 is a schematic diagram of an alternative embodiment.
FIG. 4 is a schematic diagram of the preferred embodiment of the invention.
FIG. 1 shows a conventional prior art CMOS oscillator. The circuit is powered from a VDD supply connected between terminals 10(+) and 11(-). P-channel transistor 12 and N-channel transistor 13 have their source-drain circuits coupled in series across the power supply and their gates are coupled together to provide the circuit input. The transistor drains coupled to node 14 provide the oscillator signal output. Resistor 15 provides the d-c input bias. Quartz crystal 16, in conjunction with capacitors 17 and 18, operates in the anti-resonant mode to provide signal feedback of the proper phase to support oscillation. If desired, either or both of capacitors 17 and 18 can be made variable to trim the oscillator frequency. In order to oscillate, such a circuit requires that the invertor gain exceed the losses in the feedback loop. The inverter operates as a linear amplifier. Therefore, both of the transistors will be conductive through a substantial fraction of the oscillation cycle. When the supply voltage is at the sum of thresholds, the circuit has the greatest gain. However, if the supply exceeds this value, the circuit becomes increasingly inefficient. Most of such oscillator circuits are intended for battery operation and the supply voltage, particularly for new batteries, is selected to be well above the sum of thresholds level.
FIG. 2 illustrates the nature of the invention. A voltage reference supply 20 has been added to the oscillator and a capacitor 21 employed to couple the transistor gates together. High value resistors 22 and 23 couple reference voltages to transistor 12 and 13 respectively. Capacitors 28 and 29 act as signal bypass elements so that the oscillator signals are not coupled to mirror 20. Otherwise, the oscillator circuit functions normally.
Resistor 22 couples a reference potential one P-channel threshold below VDD to the gate of transistor 12. Resistor 23 couples a reference potential one N-channel threshold above ground to the gate of transistor 13. Thus transistors 12 and 13 are optimally biased class B regardless of the value of VDD. If resistors 22 and 23 are made very high, and the gate capacitance of transistor 12 quite small, the value of capacitor 21 can be small and still serve as a coupling capacitor. Since there is no gate current, these resistors can have extremely high values.
The reference circuit 20 can be a conventional CMOS current mirror as shown designed to draw a very small operating current. Transistor 33 is designed to have higher conductance than 32. For equal currents then, less gate overdrive voltage (gate-source voltage in excess of threshold) is required by 33. The voltage drop across resistor 34 is established to cause the required decrease in overdrive. Current in 33 biases P-channel transistor 31 to establish a VT drop from its source to drain. This current is mirrored in transistor 30, which in turn biases N-channel transistor 32.
Since transistors 12 and 13 are optimally biased, the oscillator will start oscillating at VDD values well below the sum of thresholds. When oscillating at higher VDD voltages, the operating current will be controlled by the biasing circuit and can be made substantially lower than that of the prior art circuit.
It can be seen that resistors 22 and 23 will act to couple a portion of the a-c signal voltages at the gates of transistors 12 and 13 to reference supply 20. Since the current mirror in supply 20 is composed of CMOS transistors, any appreciable a-c signal will be peak rectified and can produce bias shifting. This can be avoided by connecting bypass capacitors 28 and 29 as shown across the bias supply terminals. Since resistors 22 and 23 are very large, only small bypass capacitors are needed.
FIG. 3 shows an alternative oscillator bias configuration. Resistor 22 provides bias for transistor 12. Resistor 25, which is the equivalent for resistor 15 of FIG. 1, provides the bias for transistor 13. Resistor 25 will pull the gate of transistor 13 to a bias level that just barely turns it on. This means that node 14 will operate at one N-channel threshold above ground.
FIG. 4 shows the preferred circuit of the invention. Resistor 26 has been coupled in series with the source of transistor 12 and resistor 27 has been coupled in series with the source of transistor 13. These load resistors act to reduce the transient currents in the related transistors.
Voltage reference 20 is isolated from the oscillator by means of a buffer. As described above, P-channel transistors 30 and 31 operate with N-channel transistors 32 and 33 to mirror the current flowing in resistor 34. Since resistor 34 can be made quite large and the current varies as the square of the resistor value, these currents are very small. The gate of transistor 32 will be one N-channel threshold above ground while the gate of transistor 31 will be one P-channel threshold below VDD.
Transistors 35 and 36 form an inverting buffer operated from the current mirror. Since the source of transistor 36 is coupled to the source of transistor 12, the bias supplied to transistor 12 through resistor 22 will be equal to one P-channel threshold. Resistor 25 biases the gate of transistor 13 to one N-channel threshold above its source as described in connection with FIG. 3.
Capacitor 37 has a value selected so that in conjunction with resistor 22 it couples an attenuated signal from the gate of transistor 12 to the drain of transistor 36. This attenuated signal will be peak rectified at the drain of transistor 36. The d-c thus developed will act to reduce the current in transistor 12. The buffer circuit will act to ensure that oscillation will start using class B bias. Then, once oscillation has started, the bias point is changed to reduce current drain in the oscillator transistors.
The circuit of FIG. 4 was constructed as shown. The following device geometries were used:
______________________________________ |
Transistor W/L Ratio (mils) |
______________________________________ |
12 7/0.5 |
13 8/0.35 |
30 1/0.5 |
31 1/0.5 |
32 1/0.4 |
33 2/0.4 |
35 2/0.4 |
36 0.5/0.5 |
______________________________________ |
The following component values were used:
______________________________________ |
Component Value Units |
______________________________________ |
Capacitor 21 10 picofarads |
Resistor 22 125 megohms |
Resistor 25 20 megohms |
Resistor 26 0.1 megohms |
Resistor 27 0.1 megohms |
Resistor 34 0.4 megohms |
Capacitor 37 30 picofarads |
______________________________________ |
The crystal oscillator tank circuit operated at about 32 kHz. While sum of thresholds was 1.2 volts, the circuit would oscillate at voltages down to 0.5 volt and start at 0.6 volt. The current drain with 1.5-volt supply was about 0.25 microampere which is about 5 times lower than the typical prior art circuit.
The invention has been described and a working example detailed. There are alternatives and equivalents, within the spirit and intent of the invention, that will occur to a person skilled in the art upon reading the foregoing description. Accordingly it is intended that the scope of the invention be limited only by the following claims.
Patent | Priority | Assignee | Title |
10063189, | Nov 29 2016 | STMicroelectronics (Grenoble 2) SAS | Method and device for self-biased and self-regulated common-mode amplification |
4587497, | Dec 24 1984 | CTS Corporation | Low-power low-harmonic transistor oscillator |
4710730, | Mar 20 1987 | CHAMPION TECHNOLOGIES, INC | Data clock oscillator having accurate duty cycle |
4783620, | Jul 11 1986 | Kabushiki Kaisha Toshiba | Constant voltage circuit having an operation-stop function |
4806804, | Mar 12 1986 | Deutsche ITT Industries GmbH | Mosfet integrated delay line for digital signals |
4864256, | Sep 09 1988 | SPECTRUM CONTROL, INC | Oscillator with reduced harmonics |
5184094, | Aug 16 1991 | MYCROSENSOR TECHNOLOGIES, LLC | Low power oscillator circuits |
5545941, | Jun 30 1994 | Sony Corporation | Crystal oscillator circuit |
5568093, | May 18 1995 | National Semiconductor Corporation | Efficient, high frequency, class A-B amplifier for translating low voltage clock signal levels to CMOS logic levels |
5909152, | Feb 28 1997 | Texas Instruments Incorporated | Low power CMOS crystal oscillator circuit |
6278336, | Feb 27 1998 | Texas Instruments Incorporated | Low-current oscillator with hysteresis input buffer |
6278338, | May 01 2000 | Qualcomm Incorporated | Crystal oscillator with peak detector amplitude control |
6717483, | Apr 01 2002 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Low cost voltage controlled crystal oscillator (VCXO) |
7123109, | Dec 15 2004 | MARVELL INTERNATIONAL LTD | Crystal oscillator with variable bias generator and variable loop filter |
7355489, | Feb 10 2006 | MONTEREY RESEARCH, LLC | High gain, high frequency CMOS oscillator circuit and method |
9816872, | Jun 09 2014 | Qualcomm Incorporated | Low power low cost temperature sensor |
Patent | Priority | Assignee | Title |
3887881, | |||
4095195, | Mar 25 1977 | Kabushiki Kaisha Meidensha | Low power dissipation crystal oscillator |
4307354, | Aug 22 1978 | Nippon Electric Co., Ltd. | Crystal oscillator circuit having rapid starting characteristics and a low power consumption |
4309665, | Mar 28 1979 | Hitachi, Ltd. | Complementary amplifier circuit |
JP542042, | |||
JP5553906, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Dec 15 1980 | National Semiconductor Corporation | (assignment on the face of the patent) | / | |||
Dec 15 1980 | RAPP A KARL | National Semiconductor Corporation | ASSIGNMENT OF ASSIGNORS INTEREST | 003845 | /0592 |
Date | Maintenance Fee Events |
Jul 01 1986 | ASPN: Payor Number Assigned. |
Dec 04 1986 | M170: Payment of Maintenance Fee, 4th Year, PL 96-517. |
Jul 16 1990 | M171: Payment of Maintenance Fee, 8th Year, PL 96-517. |
Sep 26 1994 | M185: Payment of Maintenance Fee, 12th Year, Large Entity. |
Oct 12 1994 | ASPN: Payor Number Assigned. |
Oct 12 1994 | RMPN: Payer Number De-assigned. |
Date | Maintenance Schedule |
Jun 07 1986 | 4 years fee payment window open |
Dec 07 1986 | 6 months grace period start (w surcharge) |
Jun 07 1987 | patent expiry (for year 4) |
Jun 07 1989 | 2 years to revive unintentionally abandoned end. (for year 4) |
Jun 07 1990 | 8 years fee payment window open |
Dec 07 1990 | 6 months grace period start (w surcharge) |
Jun 07 1991 | patent expiry (for year 8) |
Jun 07 1993 | 2 years to revive unintentionally abandoned end. (for year 8) |
Jun 07 1994 | 12 years fee payment window open |
Dec 07 1994 | 6 months grace period start (w surcharge) |
Jun 07 1995 | patent expiry (for year 12) |
Jun 07 1997 | 2 years to revive unintentionally abandoned end. (for year 12) |